Calibre 9.7: Bug Fixes in Viewer, AI, and Content Server
On April 10, 2026, version 9.7 of the open-source Calibre project was released—a cross-platform e-book manager. Built on Python and C under the GNU GPL v3.0 license, it supports Windows, macOS, and Linux. Since the 7.0 release in November 2023, the project has evolved: 8.0 in March 2025, 9.0 in January 2026. Calibre handles viewing, conversion, editing, and cataloging of books in major formats, syncs with e-ink devices, extracts metadata, and converts newspapers.
In version 8.11, an "Ask AI" tab was added to the viewer: queries to selected text via providers like Google, OpenRouter, GitHub, or local Ollama models. The feature is optional—AI code loads only after setup.
Bug Fix Details
Version 9.7 addresses non-critical bugs accumulated in previous releases. Key fixes:
- Viewer: annotations and last reading data now save correctly in EPUB and other formats.
- GitHub AI backend: improved reliability of requests and response handling.
- OpenRouter and AI: "auto" mode for logical inference no longer disables all outputs.
- Content server: fixed crash during search in reading mode; full-text search returns results without issues.
- Full-text search: refined the new interface layout.
- MTP driver on Linux: fixed rare crashes when connecting devices with large libraries (>10k books).
- Adding files: covers now set for entries without them.
These changes boost stability for mid/senior developers using Calibre in library automation pipelines or e-book API integrations.
New Features
The update introduces features for easier work with annotations and devices:
- Annotations browser: grouping by any field (author, tag, date) with sorting and filters.
- Viewer: pinch gesture support on trackpad for scaling—by default changes font size, like on touchscreen.
- Content server: full offline mode for HTTPS connections, no dependency on external resources.
Additionally, added news source Cenital by Rodrigo Pazos; improved The Week, The Age, Financial Times, Mint—faster RSS loading and metadata parsing.
Key Points
- AI Stability: GitHub and OpenRouter fixes ensure predictable queries to selected text without false disables.
- Content Server: Autonomous HTTPS mode simplifies deployment in isolated networks.
- Annotations: Grouping speeds up analysis of large collections for automated workflows.
- Cross-Platform: Linux MTP fix resolves issues with Android devices in enterprise environments.
- News Updates: New/improved sources for daily content monitoring.
The project has been developing for 11+ years, focusing on robustness without unnecessary dependencies. For developers: source code on GitHub allows forking and patching for custom needs, e.g., integration with local LLMs via Ollama API.
— Editorial Team
No comments yet.